Research On Target Detection And Recognition Based On Frequency Modulated Continuous Wave Radar

With the rise of the Internet of things and intelligent manufacturing,there is an increasing demand for target detection and recognition in many fields,such as autopilot,industrial automation,intelligent medicine,and so on.Frequency modulation continuous wave radar has attracted more and more attention in the field of target detection and recognition because of its high accuracy,small size,low cost,and strong anti-jamming ability.Compared with the ultrasonic,infrared,camera,and other sensors,Frequency modulation continuous wave radar has unique advantages in all-weather work,environmental adaptability,privacy protection,and so on.Therefore,this thesis studies the target detection and recognition method based on 77 GHz frequency modulation continuous wave radar.The main research contents are as follows:The high precision ranging method and system realization based on frequency modulation continuous wave radar is studied.Aiming at the measurement error caused by the range fence effect of frequency modulation continuous wave radar,a Zoom FFTRife joint frequency estimation algorithm is proposed to calculate the radar intermediate frequency signal frequency.The effectiveness of the proposed algorithm is verified by simulation.Based on the IWR1843 radar module,a ranging system based on frequency modulation continuous wave radar is designed to achieve high precision ranging of frequency modulation continuous wave radar.The method and system realization of multi-target detection based on frequency modulation continuous wave radar is studied.In order to solve the problem of target shadowing effect in multi-target detection of frequency modulation continuous wave radar,a two-dimensional cell average constant false alarm method with feedback mechanism is designed to reduce the effect of shadowing effect.the effectiveness of the two-dimensional cell average constant false alarm based on feedback is verified by simulation.Based on IWR1843 radar module,a multi-target detection system based on frequency modulation continuous wave radar is designed and implemented.A gesture recognition algorithm for frequency modulation continuous wave radar based on residual neural network and Vision Transformer network is proposed.Firstly,the local features of frequency modulation continuous wave radar gesture data are extracted by residual neural network,and then the key feature regions are classified and trained by using Vision Transformer network attention mechanism.The algorithm makes use of the attention mechanism of Vision Transformer network and the local feature extraction advantage of convolution neural network,and pays attention to the overall image information and key feature regions.Through the experimental verification on the frequency modulation continuous wave radar gesture data set,the results show that the proposed algorithm can effectively improve the accuracy of gesture recognition.There are 44 pictures,12 tables and 78 references in this thesis.